21

Chapter 2

Experimental setup

The current experiments are designed to examine the effect of volume fraction and Stokes number

(and equivalently the Reynolds number) at shear rates sufficiently high enough so that particle

interactions are expected to become important. Neutrally buoyant and settling particles are used

(ρp/ρ from 1 to 1.4). In this chapter the rheometer and the method used for making the torque

measurements are described (Section 2.1). Subsequently, the particles and the interstitial liquids

that were employed to create the liquid-solid mixtures are reported in Section 2.2. Finally, the

modification of the experimental apparatus for making visualizations of the flow and its procedure

is specified in Section 2.6.

2.1 Rheometer

In Figure 2.1 a schematic of the concentric-cylinder rheometer is shown. The rheometer is designed

to measure the shear stress of liquid-solid mixtures that feature relatively large particles (the order

of mm size) and are sheared at considerably shear rates. One of the difficulties that arises with high

shear rates flows is the presence of secondary flows. As shown by the review of Hunt et al. (2002),

an improper design of an experiment can result in observations that do not reflect the behavior

of the particulate flow itself. For this reason, the rheometer is designed to delay the presence of

secondary flows (Taylor vortices). This is the same experimental apparatus used by Koos (2009).

It consists of a rotational outer cylinder and an inner cylinder that consists of three parts: the top

and bottom static cylinders (fixed guard cylinders), and the middle cylinder (test cylinder) which is

instrumented to make measurements of the torque. The concentric and rotating outer cylinders are

all made of stainless steel.

The liquid-solid mixture is sheared by the rotation of the outer drum and the torque measure-

ments take place at the test cylinder. This cylinder is supported by a central shaft, which, in turn,

has ball bearings mounted to allow the free deflection of this cylinder. A set of ball bearings is also

located between the rotating outer cylinder and the fixed guard cylinders to reduce friction. The

22

h =

36

.98

cm

ro = 19.05 cm ri = 15.89 cm

Fluid injectionports

Observationports

Fixed guardcylinders

estcylinder

Rotating outercylinder

h =

36.9

8 cm

ro =

19.05 cm

ri =

15.89 cm

Fluid

injectio

n

ports

Observ

ation

ports

Fixed guard

cylinders

est

cylinder

Rotatin

g outer

cylinder

Figure 2.1: Rheometer, Couette flow device. The liquid-solid mixture is sheared by the rotationof the outer cylinder. The top and bottom cylinders are fixed. The inner middle cylinder (testcylinder) is free to rotate slightly so as to measure the torques created by the flow.

23

Property Smooth walls Rough wallsAnnulus height (hT ) 36.98 cm 36.98 cm

Test cylinder 11.22 cm 11.22 cmheightFixed guards 12.7 cm 12.7 cmheightAnnulus inner 15.89 cm 16.22 cmradius (ri)Annulus outer 19.05 cm 18.72 cmradius (ro)

Annulus gap (b) 3.16 cm 2.49 cmRatio of height 11.7 14.84to gap (hT /b)Ratio of gap to 0.166 0.133outer radius (b/ro)

Maximum rotational 14.9 rad s−1 14.9 rad s−1speed (ω)Re critical 1.6× 104 1.1× 104

Table 2.1: Rheometer properties and dimensions

fixed guards and the central shaft are equipped with seals that prevent the fluid from entering the

bearings. The fixed guard cylinders are separated from the test cylinder by a knife edge gap (0.7

mm) to prevent the particles from exiting the annulus gap but allowing the test cylinder to rotate

freely. The height of the test cylinder, H, is 11.22 cm, the inner radius of the annulus, ri, is 15.89

cm, the outer radius of the annulus, ro, is 19.05 cm, and the width of the annulus between the

cylinders when the walls are smooth, b, is 3.16 cm. The outer cylinder is driven by a belt connected

to a motor; the maximum rotational speed is ω = 14.8 rad s−1. Hence, the maximum shear rate,

γ̇ = 2ωr2o/(r2

o − r2i ), is 123 s−1. The experiment dimensions and its range of speeds are listed in

Table 2.1. Mechanical drawings of the rheometer parts can be found in the thesis of Koos (2009).

To avoid slip at the wall, the inner and outer cylinder walls are roughened by coating them with

polystyrene particles. The particles are glued to thin waterproof vinyl sticker sheets (commonly used

as aquarium backgrounds), which covered both surfaces. The glued particles are oriented randomly

and have a surface area fraction of 0.70. The decrease in gap thickness due to the rough walls is

considered when calculating the shear rate and the change in dimensions is reported in table 2.1.

Shearing the mixture by rotating the outer cylinder is preferred over shearing it by rotating

the inner one because Taylor vortices develop at lower rotational speeds in the latter case due to

centripetal forces (Taylor, 1936a,b; Wendt, 1933). Besides delaying the presence of secondary flows

by rotating the outer cylinder, the rheological measurements are made away from the top and bottom

boundary where the secondary flows develop. The contribution to the measured torque by end effects

is attenuated by increasing the ratio of height and shearing gap width, (H/b = 11.7). Further delay

24

is achieved through the increase in the ratio of gap width to outer radius (b/ro). By using a modified

gap Reynolds number (Re∗b = ρroωb/µ, where ω is the angular velocity of the outer cylinder), it is

possible to determine the critical Re∗b at which these secondary flows occur (Taylor, 1936a,b). Using

the data of Taylor, a critical modified gap Reynolds number of 1.6× 104 is found for this apparatus

when the walls are smooth and 1.1×104 when the reduction in gap width due to added roughness is

considered. The modified gap Reynolds number range for a pure fluid (no particles) in the current

experiments is 1.0× 103 ≤ Re∗b ≤ 7.2× 104, and thus some of the experiments are above the critical

Reynolds number; however, the presence of solid particles increases the effective viscosity of the flow

and hence decreases the Reynolds number.

2.2 Torque measurements

Unlike Bagnold (1954), in which the torque was not measured directly, the torque measurements in

this work are directly taken in the test section located in the middle of the inner cylinder (see Figure

2.1) without removing any possible contribution to friction or fluid. The test cylinder is allowed to

deflect rotationally due to the torque applied by the flow. The rotation of the test cylinder is opposed

by a linear spring that is connected to a static reference. The spring connects to the test cylinder

through a torque arm. The torque is measured by measuring the elongation of the spring. Different

springs with different stiffness are used depending on the torque applied by the flow. To overcome

the spring initial tension every spring is preloaded. A pulley is used to direct the preload and

reduce friction. An optical probe and a fotonic sensor (MTI 0623H and MTI KD-300, respectively)

is mounted on a static reference to measure the spring displacement. The arm that connects the

test cylinder to the spring is equipped with a mirror that is used as the moving target for the optic

probe. A sketch of the torque measurement setup is shown in Figure 2.2. The calibrations of the

optical probe and the springs are discussed in Subsection 2.2 and 2.2, respectively, followed by the

methodology for measuring the rheometer angular speed.

Optical probe calibration

The optical probe contains a set of light transmitting and light receiving fibers that are aligned

in a hemispherical configuration. Light is fed to the transmit fibers, where it exits the probe tip

and hits the target. Light that is reflected from the target is captured by the receive fibers and

transmitted to the fotonic sensor that converts the light intensity to voltage. The output voltage

is proportional to the distance between the probe tip and the target being monitored. Figure 2.3

shows a diagram of the fotonic sensor output as a function of target displacement. When the target

is in contact with the optic probe, no light is received by the fibers, giving an output signal of zero.

The voltage increases with increasing distance reaching a maximum. Past the optical peak, there is

25

Test cylinder

Torque arm Mirror

Fixed guards

Optical probe Spring

Fotonic sensor

Computer

Preload

Pulley

a) b) Figure 2.2: Torque measurement system. The outer rotating cylinder is left out of the drawingfor clarity. (a) Front view of inner cylinder. The mirror attached to the torque arm is the targetmonitored by the optical probe. (b) Side view of the inner cylinder. The optical probe and thespring are attached to a stationary frame. The spring is connected to the test cylinder through thetorque arm. The signal from the fotonic sensor is read out and stored in the computer.

26

a sensitive linear output response (Range 2). The optical probe is positioned far enough from the

moving target to make sure that the sensor output is in this range. The optical probe sensitivity

post the optical peak is 30 µm/mV . The optical probe is calibrated using a dial gage. A typical

calibration curve is shown in Figure 2.4. The curve that best fits the data is a rational polynomial

and its coefficients depend on the initial target position. The optical probe is calibrated prior to

every set of experiments and when a spring is changed. This calibration is repeated at least 10 times

and the final calibration curve is the average of these calibration curves. The fotonic sensor records

the optical probe signal for 10 seconds with a frequency sample of 10,000 Hz. The sensor output is

then averaged and coverted it to displacement using the calibration curve.

Target displacement

Voltage

outp

ut

Range 2

Optic peak

A Linear output

Figure 2.3: Diagram of an MTI KD-300 fotonic sensor output as a function of target displacement.

Springs calibration

The springs used to measure the torque were manufactured by Century springs. The calibration

of the spring is carried out by loading the spring with known masses. The masses are weighted

prior to each calibration using a scale with a resolution of 0.05 grams. The masses are attached to

the torque arm through a fishing line that runs through a pulley (see Figure 2.2). The torque arm

is connected to the spring and hence the spring can be loaded and its displacement measured for

different masses. Springs with different stiffness are used to allow a range of torques (M) between

1.3 × 10−3 ≤ M ≤ 2.7 Nm to be measured. The calibration curves for all of the springs used

are shown in Figure 2.5. The spring calibration is performed while the rheometer is filled with

27

Voltage output V0 0.5 1 1.5 2 2.5 3

Displa

cem

entcm

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Data

Fit y = !0:97x3+4:2x2+!4:4x+1:5x4+0:47x3+!1x2+0:075x+0:24 ; R

2 = 0:9998

Figure 2.4: An example of a fotonic sensor calibration curve. All the calibration curves are bestfitted by a rational polynomial. The order of the polynomial numerator and denominator is 3 and4, respectively.

water and running at different speeds. Each point represents the mean of at least 15 individually

recorded measurements; the error bars represent the standard deviation in these measurements. The

R-squared average for all the linear fits is 0.999. The lowest R-squared corresponds to the spring

with the highest slopes (and equivalently with the highest sensitivity). An initial torque is required

to start the displacement of the spring; this torque is given by the x-intercept of the calibration

curve. All the experiments are preloaded with the corresponding spring initial torque to guarantee

maximum sensitivity. The calibration curve for the spring with the highest resolution and sensitivity

is shown in Figure 2.6. The initial torque has been subtracted to show the maximum resolution of

this spring. A maximum deviation from the linear fit of 17% occurs at the lowest torques applied.The

R-squared value of this fit is 0.9988.

During the torque measurements, the displacement of the spring is also recorded with a digital

camera to account for any possible mis-calibration of the optical probe. The photos taken of the

spring are used to measure its displacement and double check the sensor output.

Angular speed measurements

The rotation of the outer cylinder is measured using a magnetic sensor and a laser tachometer.

The two devices are used to account for uncertainties in the speed measurements. For the lowest

28

Torque (Nm)0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Displa

cem

ent(c

m)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Spring 171Spring 25Spring 7Spring 23Spring 80288Spring 5169Spring 69Linear -t

Figure 2.5: Springs calibration curves. The spring constant and its sensitivity are given by the slopeof the linear fit. The error bars correspond to the standard deviation of the measurements.

Torque (Nm)0 0.005 0.01 0.015 0.02 0.025 0.03

Displa

cem

ent(c

m)

0

0.5

1

1.5

2

2.5

Slope = 71.67 cmNm

Spring 171Linear -t

Figure 2.6: Calibration curve for spring 171-b (highest sensitivity). The spring constant and itssensitivity are given by the slope of the linear fit. The error bars correspond to the standarddeviation.

29

rotational speeds, a chronometer is also used to account for the reduction of accuracy of the laser

tachometer. The magnetic sensor does not work for rotational speeds below 2.87 rad s−1. To

increase the accuracy of the laser tachometer at low speeds, 8 reflective marks were evenly spaced

and placed on the outer cylinder. Due to the high reflectiveness of stainless steel, the laser is pointed

to a black stripe painted on the outer cylinder where the reflective marks are located. Figure 2.7

shows a sketch of the rotational speed measurement system. In the previous work of Koos (2009),

Magnetic sensor

Magnet

Laser tachometer

Reflective

target

Rotating outer

cylinder

Figure 2.7: Sketch of the rotational speed measurement system. The rotational speed is measuredwith a laser tachometer and a magnetic sensor.

the angular speed was measured by counting the outer cylinder revolutions and measuring the time

with a chronometer. Measurements of the angular speed using this method were performed while

measuring the angular speed with the laser and magnet sensor to measure the uncertainty on the

previous speed measurements. The outer cylinder is driven by a belt connected to a motor; the motor

speeds are controlled via a Bardac drive that allows the gradual increase of the speed. To test that

the rheometer runs at a constant speed, rotational speed measurements for the laser, magnet sensor,

and chronometer were recorded for a period of 1.5 hours for 9 different speeds. Figure 2.8 shows

the rotational speed measurements for each instrument. The measurements are in good agreement;

the highest deviation is found at the highest rotational speeds for the measurements that used the

chronometer. The outer cylinder angular speed is shown to maintain a constant speed for all the

different speeds tested. Figure 2.9 shows the measured angular speed for each control speed tested.

Each point represents the mean of all the measurements taken from the 3 instruments. Even when

the chronometer method exhibits the biggest deviation, this deviation is no larger than 6% of the

average speed measured using the laser tachometer. Hence, the maximum uncertainty in the speed

measurements from the previous work of Koos (2009) is less than 6%.

30

Time (min)100 101 102

Rota

tionalsp

eed

(rad

s!1)

0

2

4

6

8

10

12

14

16

Figure 2.8: Measured rotational speed as a function of time for each measurement device. x , ©and 4 symbols correspond to the rotational speed measurements used using the magnetic sensor,laser tachometer, and chronometer, respectively.

Speed

1

Speed

2

Speed

3

Speed

4

Speed

5

Speed

6

Speed

7

Speed

8

Speed

9

Rot

atio

nal

spee

d(r

ads!

1)

0

2

4

6

8

10

12

14

16

Figure 2.9: The mean rotational speed measured with the three different instruments as a functionof the motor speed tested. The error bars represent the standard deviation in these measurements.

31

2.3 Error analysis

The uncertainty contributions considered for the torque measurements are:

• Uncertainty in spring constant measurement

• Uncertainty in distance measurement

• Uncertainty in optical probe calibration curve

A set of several repeated readings has been taken for each contribution. The mean and standard

deviation are calculated for each set. The estimated standard uncertainty for each contribution is

calculated as follows:

u = σ√n, (2.1)

where u is the standard uncertainty of the measurement, σ is the standard deviation, and n is the

number of measurements. Each individual standard uncertainty is then combined in terms of relative

uncertainty . The combined uncertainty is given in root mean squared (RMS) of each individual

uncertainty terms,

uT =√(u(K)

K

)2+(u(D)

D

)2+(u(V )

V

)2(2.2)

where u(K), u(D), and u(V ) are the standard uncertainty of the spring constant, distance, and

optical probe calibration, respectively. As mentioned in Subsection 2.2, the calibration curve of

the optical probe used for each experiment is the mean of at least 10 curves obtained prior to

each experiment. The distance uncertainty corresponds to the spring elongation measurements

performed to measure the torque of the flow. At least five torque measurements are taken for

each speed tested. To estimate the spring constant uncertainty, at least ten measurements of the

calibration curves are considered for each spring. The uncertainty of the rotational speed is estimated

as the standard deviation in the speed measurements. Both the laser tachometer and the magnet

sensor are used to record the outer cylinder speed for each torque measurement. The change in

suspending liquid temperature is recorded thoughout the duration of the experiments. The variation

in temperature and consequently the variation in suspending liquid density and viscosity, contributes

to the uncertainty in Stokes and Reynolds number (uSt, uRe). For these dimensionless numbers the

combined uncertainty is given by

uSt = uRe =

√(u(ω)ω

)2+(u(µ)

µ

)2+(u(ρ)

ρ

)2, (2.3)

where u(ω), u(µ), and u(ρ) are the standard uncertainty of the rotational speed, viscosity, and

density of the liquid, respectively.

32

2.4 Pure fluid torque measurements

To test the experimental method, torque measurements are performed for an aqueous-glycerine

mixture with no particles and plain water. These measurements are taken with rough walls. The

temperature of the fluid and density are monitored and considered when calculating the fluid prop-

erties. The range of gap Reynolds number tested is 3.6 × 102 ≤ Reb ≤ 8.3 × 104. The torque

measurements are compared with the theoretical results for Couette flow. Considering an infinitely

long cylindrical Couette device with the outer cylinder rotating and the inner cylinder being held

stationary, the torque applied by a laminar flow to the inner cylinder considering smooth walls is

(Schlichting, 1951)

Mi = −Mo = 4πµH ω̇r2i r

2o

r2o − r2

i

= Mlaminar, (2.4)

where H is the height of the test cylinder, µ and ρ are the viscosity and density of the fluid,

respectively, and ri and ro are the inner and outer cylinder radius. The decrease in gap thickness

due to the rough walls have been taken into account for the values of ri and ro. The effect of rough

walls is discussed later in this section. Figure 2.10 shows the torque as a function of the shear rate

for a 77% and 21% in volume aqueous-glycerine mixture and for plain water. Each point represents

the mean of 10 individually recorded measurements; the vertical error bars represent the combined

torque uncertainty. The uncertainty of the shear rate is represented by the horizontal error bars. The

theoretical results for Couette flow are also shown in Figure 2.10. For an aqueous glycerine mixture of

77%, the torque measurements are in good agreement with the theoretical laminar Couette solution.

The maximum deviation from the laminar solution is 14%. For the lower glycerine percentage of 21%

and plain water, the measured torques do not compare favorably with the torques corresponding

to laminar flow theory. For both liquids the measured torques are considerably higher than the

one predicted for laminar flow. The measured torques for water are very close in value to the ones

measured for the 21% aqueous-glycerine mixture, even though the aqueous glycerine mixture is 1.8

times more viscous than water.

Figure 2.11 shows the measured torques normalized by the theoretical torque as a function of

the gap Reynolds number defined as

Reb = ργ̇b2

µ(2.5)

The reduction in shear gap width due to the rough walls has been considered when calculating the

gap Reynolds number. For the case with the highest percentage of glycerine (and thus the lower gap

Re range), the normalized torque is close to one. The maximum uncertainty for this case occurs at

the lowest gap Reynolds number. For the case of 21% of glycerine, the normalized torque is between

4 and 15 times higher than the theoretical torque, while the normalized torque for water is between

9 and 29 times higher.

33

Shear rate s!10 20 40 60 80 100 120 140

<M

>(N

m)

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

Aqueous glycerine 77%Aqueous glycerine 21%WaterLaminar theory, considering smooth walls

Figure 2.10: Measured torque as a function of the shear rate for an aqueous-glycerine mixturewith no particles. The percentage in volume of glycerine is 77% and 21% for the aqueous-glycerinemixtures. The lines correspond to the theoretical Couette solution for a laminar flow with a viscositycorresponding to the liquid at the recorded temperatures. Continuous and dash lines correspond to77% and 21% aqueous glycerine. Dotted line is the theoretical torque for plain water.

34

Reb = ; _.b2=7102 103 104 105

M=M

lam

inar

10 -1

100

101

102

Aqueous glycerine 77%, rough walls

Aqueous glycerine 21%, rough walls

Water, rough walls

Laminar theory, considering smooth walls

Figure 2.11: Normalized torque as a function of the gap Reynolds number for an aqueous-glycerinemixture of 77% and 21% and plain water with no particles.

To compare the current results with the work of Taylor (1936a), Figure 2.12 shows the normalized

torques as a function of the modified gap Reynolds number used by Taylor, defined as

Re∗b = ρωrob

µ. (2.6)

Based on the work of Taylor (1936a,b) and considering the reduction in gap due to the rough walls,

the critical gap Reynolds number at which the flow becomes unstable for the current apparatus

geometry is 1.1× 104 (for rough walls). The range of gap Reynolds number corresponding to plain

water and 21% aqueous glycerine is between 4.6 × 103 ≤ Re ≤ 8.31 × 104. Therefore, most of

these experiments are above the critical Reynolds number. Figure 2.12 also shows the normalized

torques from Koos (2009) performed with smooth walls. Comparing the smooth and rough walls

measurements, the ratio of torques for the latter deviates from the laminar behavior at modified gap

Reynolds where the smooth walls show laminar behavior. This suggests that the presence of rough

walls can lower the critical Reynolds number.

Figure 2.13 shows a comparison between the current measurements and the data of Bagnold

(1954); Taylor (1936a) and Wendt (1933) for pure fluid torque measurements normalized with the

torque predicted from laminar theory. The normalized torques are presented as a function of the

modified gap Reynolds number (Re∗b) used by Taylor (1936a) and defined as equation 2.6. All of

35

Re$b = ;!rob=7102 103 104 105

M=M

lam

inar

10 -1

100

101

102

A Critical gap Re

Aqueous glycerine 75%, smooth walls from Koos (2009)

Aqueous glycerine 68%, smooth walls from Koos (2009)

Aqueous glycerine 77%, rough walls

Aqueous glycerine 21%, rough walls

Water, rough walls

Laminar theory, considering smooth walls

Figure 2.12: Closed symbols: normalized torques as a function of the modified gap Reynolds numberfor pure fluids measured with rough walls. Open symbols: normalized torques for pure fluid measuredwith smooth walls from Koos (2009). Vertical dashed line represents the critical modified Re basedon the work of Taylor (1936a) and considers the gap width for rough walls.

36

these studies were performed with smooth walls. The shear gap width to outer radius ratio for the

current experiments is smaller (b/ro = .133) than the corresponding ratio for the data of Taylor

(b/ro = .15 and b/ro = .21), Wendt (b/ro = .15), and Bagnold (b/ro=.19). A smaller ratio leads to

lower critical gap Reynolds number. The normalized torques for the lower range of gap Re compares

favorably with the current results; however, a deviation from laminar theory occurs at a lower

Reynolds number than the one predicted using the data of Taylor (1936a). Cadot et al. (1997); Lee

et al. (2009) and van den Berg et al. (2003) studied the effect of rough boundaries on Taylor-Couette

flow, where the liquid was sheared by the rotation of the inner cylinder. These studies show that the

presence of rough walls does not affect the laminar transition. A lower critical Re might be due to a

smaller ratio in the annulus height to shear gap width. For the experiments of Taylor, this ratio was

hT /b = 99 and hT /b = 141, and for the experiments of Wendt, hT /b = 28, which is considerably

higher than the ratio for the current experiments (hT /b = 14.9). Moreover, when the flow is driven

by the rotation of the outer cylinder, Taylor (1936a) observed that the transition occurred at a

range of Reynolds number rather than at an specific one. Experiments on Taylor-Couette flow with

rough walls and inner cylinder rotating indicate that the presence of rough walls does not affect the

instability of Taylor vortex flow for low Reynolds numbers, but it does intensify the turbulent Taylor

vortex flow at Reynolds numbers above the critical one (Lee et al., 2009). This rough walls effect

leads to higher measured torques. The measured torque for pure fluid for the current experiments

is consistent with these findings.

To verify the Newtonian behavior of the liquids used for these experiments, viscosity measure-

ments were performed using a strain-controlled rheometer (TA instruments, ARES-RFS, Rheomet-

rics fluid Spectrometer). The results obtained using the rheometric fluid spectrometer indicate that

the liquids used are indeed Newtonian.

2.5 Particles

Two type of particles are used in these experiments: polystyrene elliptical cylinders and polyester

scalene ellipsoids. The properties of these particles are summarized in table 2.2. The diameter

reported in table 2.2 is the equivalent spherical particle diameter. These particles are the same

particles used by Koos et al. (2012). The particles vary in size, shape, and density. These

variations lead to different values for the random loose and random-close packing. The values used

for these packings were obtained from the measurements done by Koos (2009).

Polystyrene: density ratio = 1 and 1.05

Figure 2.14 shows the polystyrene particles used in the current experiments, which are elliptical

cylinders. The specific gravity for these particles is 1.05 and they are neutrally buoyant in an aqueous

37

Re$b = ;!b=7102 103 104 105

<M

>=M

lam

inar

10 -1

100

101

102

Bagnold, b=ro = 0:19Taylor b=ro = :15Taylor b=ro = :21Wendt b=ro = :15, -xedWendt b=ro = :15 corotatingAqueous glycerine 77% b=ro = :13Aqueous glycerine 21% b=ro = :13Water b=ro = :13

Figure 2.13: Pure fluid torque measurements normalized with laminar Couette flow as a functionof modified gap Reynolds number. Comparison between pure fluid data of Bagnold (1954); Taylor(1936a) and Wendt (1933).

Property Polystyrene Polyester

diameter, d (mm) 3.34 2.93gap width smooth walls

diameter , bs/d 9.46 10.79

gap width rough wallsdiameter , b/d 7.46 8.5

particle density, ρp (kg/m3) 1050 1400

liquid density, ρ (kg/m3) 1000-1050 1000-1160.5shape elliptical cylinders ellipsoids

sphericity, ψ 0.7571 0.9910RLP, φRLP 0.553 0.593RCP, φRCP 0.663 0.65

Young’s modulus, E (MPa) 3000 2800

Yield Strength, Y (MPA) 40 55Poisson’s ratio ν 0.34 0.39

Table 2.2: Particles properties

38

glycerine mixture of 21% glycerine. Koos (2009) measured the particle diameter and lengths of 50

particles. The sample had an average small diameter dsmall = 2.08 mm, large diameter dlarge = 2.92,

and length l = 3.99 mm. The particle length was found to be bimodal, whereas the diameters were

found to be unimodal. By measuring the displaced volume of 1000 particles and considering that

the volume of each particle is

Vp = π

4 dsmalldlargel.

Koos (2009) found an equivalent sphere diameter of d = 3.35 mm. By weighing the same sample

and considering a particle density of ρp = 1050 kg/m3, an equivalent sphere diameter of d = 3.34

was found. Both measurements were in agreement and the particles are taken to have an equivalent

spherical diameter of d = 3.34 ± 0.02 mm. Polystyrene particles are the ones used to roughen the

Figure 2.14: Polystyrene particles. Specific gravity of 1.05 and have an equivalent spherical particlediameter of 3.34± 0.02 mm. Reference shown in cm.

rheometer walls. Figure 2.15 shows a picture of the surface roughness of the walls using polystyrene

particles.

Polyester: density ratio=1.2 and 1.4

The polyester particles used in the current experiments are shown in Figure 2.16. These particles

have a specific gravity of 1.4 and a shape of scalene ellipsoids. Koos (2009) found an equivalent

39

Figure 2.15: Rheometer walls surface roughness formed by polystyrene particles. The reference is ininches.

spherical diameter of d = 2.93± 0.02 mm. The rough walls used in these experiments are the same

Figure 2.16: Polyester particles. Specific gravity of 1.4 and have an equivalent spherical particlediameter of 2.93± 0.02 mm. Reference shown in cm.

as the ones used for the polystyrene particles.

40

2.6 Visualizations

Visualizations of the flow are performed by replacing the top fixed guard and test cylinder with a

transparent acrylic cylinder. This cylinder has the same radius as the inner cylinder: r = 15.89 cm

and it is 30 cm in length. The surface of the visualization cylinder is roughened using polystyrene

particles. A window of the same length of the cylinder height and 3.8 cm in width is left smooth

and used for visualization purposes. Figure 2.17 shows a sketch of the visualization setup. A digital

camera is located inside the visualization cylinder to record the flow. The flow is illuminated with

halogen lights located on top. An acrylic ring with a width slightly smaller than the gap shear

width is put on the rim of the visualization cylinder to work as an end cap. For the visualization of

NOTE : THIS IS PRURELY AN IMPRESSION OF THE ASSMEBLEDDEVICE. DO NOT SCALE OR USE AS AN ASSEBLY GUIDE.

REVISIONS

REV. DESCRIPTION DATE APPROVED

DRAWN BY

THIRD ANGLE PROJECTION

MACHINE FINISH

63GEN. TOL.0.X 0.020"0.XX 0.010"0.XXX 0.005"ANGLE 0.5

TITLE

MATERIAL

DATE

PROJECT

COMPLETE ASSEMBLY

DIM. INSCALE

DWG. NO.

SHEET

REV.

ZONE

ERIN KOOS X4224

N/A

NASA COAXIAL RHEOMETER

INCHES 11/10/05

01030 3

2 OF 21:3

1. ADDED FILL PORT BOSS AND 05/28/02 DRAIN PORT TAPPING2. REBUILD USING NEW PARTS 11/10/05

NOTE : THIS IS PRURELY AN IMPRESSION OF THE ASSMEBLEDDEVICE. DO NOT SCALE OR USE AS AN ASSEBLY GUIDE.

REVISIONS

REV. DESCRIPTION DATE APPROVED

DRAWN BY

THIRD ANGLE PROJECTION

MACHINE FINISH

63GEN. TOL.0.X 0.020"0.XX 0.010"0.XXX 0.005"ANGLE 0.5

TITLE

MATERIAL

DATE

PROJECT

COMPLETE ASSEMBLY

DIM. INSCALE

DWG. NO.

SHEET

REV.

ZONE

ERIN KOOS X4224

N/A

NASA COAXIAL RHEOMETER

INCHES 11/10/05

01030 3

2 OF 21:3

1. ADDED FILL PORT BOSS AND 05/28/02 DRAIN PORT TAPPING2. REBUILD USING NEW PARTS 11/10/05

Digital camera

Bottom fixed guard

Outer rotating cylinder

Acrylic transparent

cylinder

Figure 2.17: Visualization cylinder setup. The walls of the inner transparent cylinder are roughenedwith polystyrene particles except from the visualization window (3.8× 30 cm)

the flow of polystyrene particles, 20% percent of these particles were painted on one side to better

track the flow. The painted particles did not exhibit difference in density. Figure 2.18 shows a

sample of painted and not painted polystyrene particles immersed in water. The mixture seems to

be evenly mixed. Because of the replacement of the test cylinder with the fixed acrylic cylinder, the

visualizations of the flow cannot be performed simultaneously with the torque measurements.

41

Figure 2.18: Sample of polystyrene particles used for visualization purposes. The particles areimmersed in water. The mix of painted particles are evenly mixed with the non-painted ones, andthus the paint does not seem to have a considerable influence on particles density.